Google AI Mode: Official Site or Local Distributor?
Google is testing an official site section in AI Mode. In export markets that rarely means the manufacturer wins, and sometimes it should not.
Read moreAssistants and AI search
How ChatGPT, Claude, Gemini, Perplexity and Google AI Mode answer, cite and change, and what each update means for a B2B supplier.
18 articles
ChatGPT Plugins for Brands Are Not Organic Visibility
Custom GPTs are retiring, Claude has a plugin directory and OpenAI now ranks plugins in chat. Why a brand plugin and being named are two separate numbers.
Read more
The Universal Commerce Protocol, Explained for B2B Suppliers
A protocol, a cart and an agent arrived under one headline. Only one of them already matters to a supplier selling into Europe, and it is not the checkout.
Read more
ChatGPT Ads Skip Paid Plans. Does Your B2B Buyer See Them?
ChatGPT ads run only on Free and Go. Business and Enterprise workspaces never show them, so the buyer with the budget may be the one person your campaign cannot reach.
Read more
Measuring Your Brand in Google AI Mode: What You Can See
Linked, named or clicked: three different events, and only one comes with official numbers. What Search Console shows, what sampling adds, where we stop.
Read more
The Supplier Search That Happens Behind the Login
Procurement suites now have agents that propose suppliers to buyers. Nobody outside can see what they say, but what they read is largely yours to fix.
Read more
Opting Out of Google's AI Answers Rarely Helps a B2B Site
Google now lets any site leave AI Overviews and AI Mode. For a supplier selling to engineers and buyers, the switch usually guards very little worth guarding.
Read more
What Search Console's AI Report Actually Counts
Search Console now reports on AI Overviews and AI Mode for every site. It counts something real. It is not your AI visibility, and the gap is worth understanding.
Read more
An Ad Beneath the Answer Is Not in the Answer
ChatGPT now carries ads in 31 European countries. Being named by the model and being shown beneath it are two different events, and they should never share a figure.
Read more
The Same Technical Question, Four Different AI Answers
Ask four assistants who can machine a part to your spec and you get four different supplier lists. The reasons are structural, and testable in an afternoon.
Read more
AI Overviews or AI Assistants: What to Measure First
One answer sits above the search results, the other happens inside a chat. They are not interchangeable, and a good score in one predicts nothing about the other.
Read more
Gemini Is Not Google AI Mode, and Neither Is AI Overviews
Three Google surfaces get confused constantly, and the confusion sells the wrong tool. What actually separates them, and which one PSentry can see.
Read more
Brand Monitoring in Claude: The Assistant Marketers Skip
Claude answers many B2B questions from what it learned in training, not from the live web. That changes what you can measure, and how you check it.
Read more
How to Track Your Brand in Gemini, and What Gemini Is Not
Gemini is an assistant, a model family, and the engine behind Google's search summaries. Only one of those is what we measure. How to check the right one.
Read more
Perplexity Shows Its Sources. Tracking a Brand Is Still Hard
It is the only AI platform that prints the pages behind its answer. That makes it the best diagnostic instrument you have, and a poor proxy for everything else.
Read more
Why ChatGPT, Claude, Gemini and Perplexity Disagree About You
Ask the four big AI assistants the same buying question and you often get four different brand lists. Watching only one of them measures a quarter of your visibility.
Read more
How to Track Your Brand in ChatGPT for Free, and Where It Breaks
A step-by-step manual method to see whether AI assistants name your brand, the mistake that makes most DIY tests worthless, and the point where it collapses.
Read more
Brand Monitoring in ChatGPT: How to Actually Check
Asking ChatGPT what it knows about your brand always gives a reassuring answer, and it measures nothing. Here is the test that does.
Read more